Implementation:PrefectHQ Prefect Pydantic AI Agent Setup

Metadata
Sources	Prefect, pydantic-ai Agents
Domains	AI_Agents, LLM
Last Updated	2026-02-09 00:00 GMT

Overview

Concrete wrapper for configuring pydantic-ai Agent instances with tools and structured output for Prefect AI workflows.

Description

This Wrapper Doc covers the pydantic-ai Agent class as used in Prefect examples. The Agent is configured with a model identifier, output_type (Pydantic model), deps_type (dependency type for runtime context), tools (callable functions), and system_prompt. This is an external library API (pydantic-ai) used within Prefect workflows.

Code Reference

Repository: https://github.com/PrefectHQ/prefect
File: examples/ai_data_analyst_with_pydantic_ai.py (L182-195)
Signature:

agent = Agent(
    "openai:gpt-4o",
    name="data-analyst-agent",
    output_type=DataAnalysis,
    deps_type=pd.DataFrame,
    tools=[calculate_statistics, detect_anomalies, get_column_info],
    system_prompt=(
        "You are an expert data analyst. Analyze the provided dataset..."
    ),
)

Import: from pydantic_ai import Agent, RunContext
External Reference: https://ai.pydantic.dev/agents/

I/O Contract

Inputs

model (str, required) — LLM model identifier like "openai:gpt-4o"
output_type (type, optional) — Pydantic model for structured output
deps_type (type, optional) — Dependency type
tools (list[Callable], optional) — Tool functions
system_prompt (str, optional) — Agent instructions
name (str, optional) — Agent name

Outputs

Agent[DepsType, OutputType] — Configured agent instance ready for execution

Usage Example

from pydantic_ai import Agent, RunContext
from pydantic import BaseModel, Field
import pandas as pd

class DataAnalysis(BaseModel):
    summary: str
    key_findings: list[str] = Field(min_length=3, max_length=5)
    recommendations: list[str] = Field(min_length=3, max_length=5)
    columns_analyzed: list[str]

def calculate_statistics(ctx: RunContext[pd.DataFrame], column: str) -> dict:
    df = ctx.deps
    return df[column].describe().to_dict()

agent = Agent(
    "openai:gpt-4o",
    output_type=DataAnalysis,
    deps_type=pd.DataFrame,
    tools=[calculate_statistics],
    system_prompt="You are an expert data analyst.",
)

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment